Can L2 learners acquire native-like typicality representation in categorization?